Microsoft Cluster Server (MSCS) on Windows

This section discusses both Mode 1 and Mode 2 configuration in Microsoft Cluster Server. Sites using Microsoft Failover Cluster Manager, Veritas Cluster Server or Novell Cluster Services should jump to the following sections.

This section assumes that you have an already installed and working clustered printing environment.

Mode 1 - Clustering at the Print Provider layer

The PaperCut Print Provider is the component that integrates with the print spooler service and provides information about the print events to the PaperCut Application Server. At a minimum, in a cluster environment, the Print Provider component needs to be included and managed within the cluster group. The Application Server component (The Standard Install option in the installer) is set up on an external server outside the cluster. Each node in the cluster is configured to report back to the single application server using XML web services over TCP/IP.

Step 1 - Application Server (Primary Server) Setup

Install the Application Server component (Standard Install option) on your nominated system. This system will be responsible for providing PaperCut NG's web based interface and storing data. In most cases this system will not host any printers and is dedicated to the roll of hosting the PaperCut Application Server. It may be one of the nodes in the cluster; however a separate system outside the cluster is generally recommended. An existing domain controller, member server or file server will suffice.

Step 2 - Installing the Print Provider components on each node

The Print Provider component needs to be separately installed on each node involved in the print spooler cluster. This is done by selecting the Secondary Print Server option in the installer. Follow the secondary server set up notes as detailed in Chapter 14, Configuring Secondary Print Servers and Locally Attached Printers . Take care to define the correct name or IP address of the nominated application server set up in step 1.

Step 3 - Decouple service management from nodes

By default the Print Provider component is installed under the management of the node. To hand over management to the cluster, the service start-up type needs to be set to manual. On each node navigate to Control PanelAdministrative ToolsServices, locate the PaperCut Print Provider service. Stop the service and set the start-up type to Manual. Repeat for each node in the cluster.

Stopping the service and setting to Manual startup

Figure 23.1. Stopping the service and setting to Manual startup

Step 4 - Adding the Print Provider service as a resource under the print spooler's cluster group

  1. Open the Cluster Administrator.

  2. Right-click on the cluster group hosting the spooler service and select NewResource.

    Creating a new cluster resource

    Figure 23.2. Creating a new cluster resource

  3. In the new resource wizard, enter a name of PaperCut Print Provider and select a resource type of Generic Service. Click Next.

  4. Click Next at Possible Owners.

  5. Ensure that the Print Spooler Service resource is set as a required dependency, then click Next.

  6. On the Generic Service Parameters page, enter a service name of PCPrintProvider and ensure the Use Network Name for computer name option is checked. Click Next.

    Cluster service parameters configuration

    Figure 23.3. Cluster service parameters configuration

  7. Click Finish at the Registry Replication page.

Step 5 - Shared active job spool

To ensure the state of jobs currently active (e.g. held in a hold/release queue) are not lost during a failover event, PaperCut NG is able to save job state in a shared drive/directory. If a shared disk resource is available and can be added to the cluster resource, PaperCut NG can use this to host a shared spool directory to ensure no active job state is lost.

  1. Add a shared drive to the cluster resource. e.g. (Q: drive). It is advisable to use the same drive as used for the shared print spool directory.

  2. Create a directory in this drive called PaperCut NGSpool

  3. Create a sub-directory in PaperCut NGSpool called activejobs

  4. On each node, edit the file: [app-path]/providers/print/win/print-provider.conf and add a line pointing to the shared active job spool directory:

        ActiveJobsSpoolDir=Q:\PaperCut NG\Spool\activejobs\
    

    Change the drive letter as appropriate.

    Important

    If running an "Active-Active" print cluster you must use a separate active job spool directory for each node. To configure this use the %service-name% value in the ActiveJobsSpoolDir setting. The %service-name% value is replaced by the service name of the running PaperCut Print Provider instance. e.g.

        ActiveJobsSpoolDir=Q:\PaperCut NG\Spool\%service-name%\activejobs\
    

Step 6 - Bring up all cluster resources and test

Perform operations to verify that:

  1. Print jobs log as expected.

  2. No error message appear in the Print Providers text log located at: C:\Program Files\PaperCut NG\providers\print\win\print-provider.log on each node.

Active/Active Clustering - Special notes regarding multiple virtual servers

On large networks it is common to distribute load by hosting print spooler services under two or more virtual servers. For example, two virtual servers may each host half of the organization's printers and hence sharing the load. This is sometimes referred to as Active/Active clustering - all be it not an entirely correct term, as the print spooler is still running in Active/Passive.

Virtual servers cannot share the same service on any given node. For this reason if the virtual servers share nodes, you'll need to manually install the PaperCut Print Provider service a second time under a different name. This can be done via the command line as follows:

   cd C:\Program Files\PaperCut NG\providers\print\win
   pc-print.exe PCPrintProvider2 /install
                    

The argument proceeding /install is the unique name to assign to the service. The recommended procedure is to suffix the standard service name with a sequential number.

Mode 2 - Clustering at all application layers

Mode 2 implements failover clustering at all of PaperCut NG's Service Oriented Architecture software layers, including:

  • Clustering at the Print monitoring layer

  • Clustering at the Application Server layer

  • Optional clustering at the database layer

Mode 2 builds upon Mode 1 by introducing failover (Active/Passive) clustering in the Application Server layer. This involves having an instance of the application server on each of the cluster nodes. When one node fails, the other automatically takes over the operation. Both instances use a share data source in the form of an external database (see Chapter 19, Deployment on an External Database (RDBMS)). Large sites should consider using a clustered database such as Microsoft SQL Server.

This section assumes that you have an already installed and working clustered printing environment.

Step 1 - Application Server Installation

On one of the cluster's nodes, install the PaperCut Application Server component by selecting the Standard Install option in the installer. Follow the setup wizard and complete the process of importing all users into the system.

Step 2 - Convert the system over to an external database

The system needs to be configured to use an external database as this database will be shared between both instances of the application server. Convert the system over to the required external database by following the procedure detailed in Chapter 19, Deployment on an External Database (RDBMS). The database may be hosted on another system, or inside a cluster. As per the external database setup notes, reference the database server by IP address by entering the appropriate connection string in the server.properties file.

Step 3 - Setup of 2nd Node

Repeat steps 1 and 2 on the second cluster node.

Step 4 - Decouple service management from the nodes

By default the PaperCut Application Server component is installed under the management of the node. It needs to be managed inside the cluster, so the service's start-up type should be set to manual. On each node navigate to Control PanelAdministrative ToolsServices locate the PaperCut Application Server. Stop the service and set its start-up type to Manual. Repeat this on both nodes.

Step 5 - Create a new cluster group

The PaperCut Application Server should be designated to run inside its own cluster group. Create a new cluster group containing the two nodes. Add an IP Resource and a Network Name resource. Give the network name resource an appropriate title such as PCAppSrv.

The need for a new cluster group is not hard and fast. It is however recommended as it gives the most flexibility in terms of load balancing and minimizes the potential for conflicts.

Step 6 - Adding the PaperCut Application Service as a resource managed under the new cluster group.

  1. Open the Cluster Administrator.

  2. Right-click on the cluster group hosting the spooler service and select NewResource.

  3. In the new resource wizard, enter a name of PaperCut Application Server and select a resource type of Generic Service. Click Next.

  4. Click Next at Possible Owners page.

  5. Click Next at Dependency page.

  6. On the Generic Service Parameters page, enter a service name of PCAppServer and ensure the Use Network Name for computer name option is checked. Click Next.

  7. Click Finish at the Registry Replication page.

Step 7 - Bring the cluster group online

Right-click on the cluster group and select Bring online. Wait until the application server has started, then verify that you can access the system by pointing a web browser to :

    http://[Virtual Server Name]:9191/admin 
                    

Login, and perform some tasks such as basic user management and User/Group Synchronization to verify the system works as expected.

Step 8 - Set up the Print Provider layer

Interface the PaperCut Print Provider layer with the clustered spooler service by following the same setup notes as described for Mode 1. The exception being that the IP address of the application server will be the IP address assigned to the Virtual Server assigned in step 5.

Step 9 - Client Configuration

The client and release station programs are located in the directories:

  • [app-path]/client/

  • [app-path]/release/

These directories contain configuration files that instruct the client to the whereabouts of the server. The IP address and the server name in the following set of files will need to be updated to the Virtual Server's details (Name and IP address):

  • [app-path]/client/win/config.properties

  • [app-path]/client/linux/config.properties

  • [app-path]/client/mac/PCClient.app/Contents/Resources/config.properties

  • [app-path]/release/connection.properties

Edit the files using Notepad or equivalent and repeat this for each node. Also see the section called “Client/Workstation Configuration”.

Step 10 - Test

Mode 2 setup is about as complex as it gets! Take some time to verify all is working and that PaperCut NG is tracking printing on all printers and all virtual servers.

Advanced: Load distribution and independent groups

It is possible to split the two application layers (Resources) into two separate Cluster Groups:

  • Group 1: Containing only the PaperCut Application Server service.

  • Group 2: Containing the PaperCut Print Provider and Print Spooler services. These services are dependent and hence must be hosted in the same group.

Separating these resources into to groups allows you to set up different node affinities so the two groups usually run on separate physical nodes during normal operation. The advantage is that the load is spread further across the systems and a failure in one group will not necessarily fail-over the other.

To make this change after setting up the single group Mode 2 configuration:

  1. Change the ApplicationServer= option in [app-path]/providers/print/win/print-provider.conf on each physical node to the IP or DNS name of the virtual server.

  2. Create a new group called PaperCut Application Server Group.

  3. Set the Preferred owners of each group to different physical nodes.

  4. Restart or bring on line each group, and independently test operation and operation after fail-over.

Clustering Tips

Tip

Take some time to simulate node failure. Monitoring may stop for a few seconds while the passive server takes over the role. Simulating node failure is the best way to ensure both sides of the Active/Passive setup is configured correctly.

It is important that the version of PaperCut NG running on each node is identical. Ensure that any version updates are applied to all nodes so versions are kept in sync.

The PaperCut NG installation sets up a read-only share exposing client software to network users. If your organization is using the zero-install deployment method, the files in this share will be accessed each time a user logs onto the network. Your network may benefit from exposing the contents of this share via a clustered file share resource.

Additional configuration steps to support Web Print

By default the Application Server will look in [app-path]\server\data\web-print-hot-folder for Web Print files. This location is generally only available on one node in the cluster. To support Web Print in a cluster you will need to add a Shared Folder on the Shared Storage in your cluster. This can be done on the same disk that the spool files reside and the Print Provider point to.

To change this location you will need to use the Config Editor and modify the web-print.hot-folder key.

  1. Add a Shared Folder on the Shared Storage, an example would be E:\web-print-hot-folder and share it as \\clustername\web-print-hot-folder\.

  2. Log in to the PaperCut NG administration console, naviate to OptionsConfig Editor

  3. Modify web-print.hot-folder to E:\web-print-hot-folder

  4. Map your selected network drive on the Web Print Sandbox machine to \\clustername\web-print-hot-folder\

  5. Add all relevant printer queues from \\clustername to the Web Print Sandbox server.